CNNs have demonstrated superior performance in medical image segmentation. To overcome the limitation of only using local receptive field, previous work has attempted to integrate Transformers into convolutional network components such as encoders, decoders, or skip connections. However, these methods can only establish long-distance dependencies for some specific patterns and usually neglect the loss of fine-grained details during downsampling in multi-scale feature extraction. To address the issues, we present a novel hybrid Transformer network called FocalTransNet. specifically, we construct a focal-enhanced (FE) Transformer module by introducing dense cross-connections into a CNN-Transformer dual-path structure and deploy the FE Transformer throughout the entire encoder. Different from existing hybrid networks that employ embedding or stacking strategies, the proposed model allows for a comprehensive extraction and deep fusion of both local and global features at different scales. Besides, we propose a symmetric patch merging (SPM) module for downsampling, which can retain the fine-grained details by stablishing a specific information compensation mechanism. We evaluated the proposed method on four different medical image segmentation benchmarks. The proposed method outperforms previous state-of-the-art convolutional networks, Transformers, and hybrid networks. The code for FocalTransNet is publicly available at https://github.com/nemanjajoe/FocalTransNet.

The respiratory system depends on complex biomechanical processes to enable gas exchange. The mechanical properties of the lung parenchyma, airways, vasculature, and surrounding structures play an essential role in overall ventilation efficacy. These complex biomechanical processes however are significantly altered in chronic obstructive pulmonary disease (COPD) due to emphysematous destruction of lung parenchyma, chronic airway inflammation, and small airway obstruction. Recent advancements computed tomography (CT) and magnetic resonance imaging (MRI) acquisition techniques, combined with sophisticated image post-processing algorithms and deep neural network integration, have enabled comprehensive quantitative assessment of lung structure, tissue deformation, and lung function at the tissue level. These methods have led to better phenotyping, therapeutic strategies and refined our understanding of pathological processes that compromise pulmonary function in COPD. In this review, we discuss recent developments in imaging and image processing methods for studying pulmonary biomechanics with specific focus on clinical applications for chronic obstructive pulmonary disease (COPD) including the assessment of regional ventilation, planning of endobronchial valve treatment, prediction of disease onset and progression, sizing of lungs for transplantation, and guiding mechanical ventilation. These advanced image-based biomechanical measurements when combined with clinical expertise play a critical role in disease management and personalized therapeutic interventions for patients with COPD.

The nnUNet segmentation framework adeptly adjusts most hyperparameters in training scripts automatically, but it overlooks the tuning of internal hyperparameters within the segmentation network itself, which constrains the model's ability to generalize. Addressing this limitation, this study presents a novel Self-Adaptive Convolution Module that dynamically adjusts the size of the convolution kernels depending on the unique fingerprints of different datasets. This adjustment enables the MSA2-Net, when equipped with this module, to proficiently capture both global and local features within the feature maps. Self-Adaptive Convolution Module is strategically integrated into two key components of the MSA2-Net: the Multi-Scale Convolution Bridge and the Multi-Scale Amalgamation Decoder. In the MSConvBridge, the module enhances the ability to refine outputs from various stages of the CSWin Transformer during the skip connections, effectively eliminating redundant data that could potentially impair the decoder's performance. Simultaneously, the MSADecoder, utilizing the module, excels in capturing detailed information of organs varying in size during the decoding phase. This capability ensures that the decoder's output closely reproduces the intricate details within the feature maps, thus yielding highly accurate segmentation images. MSA2-Net, bolstered by this advanced architecture, has demonstrated exceptional performance, achieving Dice coefficient scores of 86.49\%, 92.56\%, 93.37\%, and 92.98\% on the Synapse, ACDC, Kvasir, and Skin Lesion Segmentation (ISIC2017) datasets, respectively. This underscores MSA2-Net's robustness and precision in medical image segmentation tasks across various datasets.

Abnormality detection in medical imaging is a critical task requiring both high efficiency and accuracy to support effective diagnosis. While convolutional neural networks (CNNs) and Transformer-based models are widely used, both face intrinsic challenges: CNNs have limited receptive fields, restricting their ability to capture broad contextual information, and Transformers encounter prohibitive computational costs when processing high-resolution medical images. Mamba, a recent innovation in natural language processing, has gained attention for its ability to process long sequences with linear complexity, offering a promising alternative. Building on this foundation, we present SpectMamba, the first Mamba-based architecture designed for medical image detection. A key component of SpectMamba is the Hybrid Spatial-Frequency Attention (HSFA) block, which separately learns high- and low-frequency features. This approach effectively mitigates the loss of high-frequency information caused by frequency bias and correlates frequency-domain features with spatial features, thereby enhancing the model's ability to capture global context. To further improve long-range dependencies, we propose the Visual State-Space Module (VSSM) and introduce a novel Hilbert Curve Scanning technique to strengthen spatial correlations and local dependencies, further optimizing the Mamba framework. Comprehensive experiments show that SpectMamba achieves state-of-the-art performance while being both effective and efficient across various medical image detection tasks.

Medical image retrieval is essential for clinical decision-making and translational research, relying on discriminative visual representations. Yet, current methods remain fragmented, relying on separate architectures and training strategies for 2D, 3D, and video-based medical data. This modality-specific design hampers scalability and inhibits the development of unified representations. To enable unified learning, we curate a large-scale hybrid-modality dataset comprising 867,653 medical imaging samples, including 2D X-rays and ultrasounds, RGB endoscopy videos, and 3D CT scans. Leveraging this dataset, we train M3Ret, a unified visual encoder without any modality-specific customization. It successfully learns transferable representations using both generative (MAE) and contrastive (SimDINO) self-supervised learning (SSL) paradigms. Our approach sets a new state-of-the-art in zero-shot image-to-image retrieval across all individual modalities, surpassing strong baselines such as DINOv3 and the text-supervised BMC-CLIP. More remarkably, strong cross-modal alignment emerges without paired data, and the model generalizes to unseen MRI tasks, despite never observing MRI during pretraining, demonstrating the generalizability of purely visual self-supervision to unseen modalities. Comprehensive analyses further validate the scalability of our framework across model and data sizes. These findings deliver a promising signal to the medical imaging community, positioning M3Ret as a step toward foundation models for visual SSL in multimodal medical image understanding.

Diagnosing sarcoidosis remains challenging. Histology findings and a variable clinical presentation can mimic other infectious, malignant, and autoimmune diseases. This review synthesizes current evidence on histopathology, sampling techniques, imaging modalities, and biomarkers and explores how emerging 'omics' and artificial intelligence tools may sharpen diagnostic accuracy. Within the typical granulomatous lesions, limited or 'burned-out' necrosis is an ancillary finding, which can be present in up to one-third of sarcoid biopsies, and demands a careful differential diagnostic work-up. Endobronchial ultrasound-guided transbronchial needle aspiration of lymph nodes has replaced mediastinoscopy as first-line sampling tool, while cryobiopsy is still under validation. Volumetric PET metrics such as total lung glycolysis and somatostatin-receptor tracers refine activity assessment; combined FDG PET/MRI improves detection of occult cardiac disease. Advanced bronchoalveolar lavage (BAL) immunophenotyping via flow cytometry and serum, BAL, and genetic biomarkers show to correlate with inflammatory burden but have low diagnostic value. Multi-omics signatures and Positron Emission Tomography with Computer Tomography radiomics, supported by deep-learning algorithms, show promising results for noninvasive diagnostic confirmation, phenotyping, and disease monitoring. No single test is conclusive for diagnosing sarcoidosis. An integrated, multidisciplinary strategy is needed. Large, multicenter, and multiethnic studies are essential to translate and validate data from emerging AI tools and -omics research into clinical routine.

The emergence of powerful, general-purpose omnimodels capable of processing diverse data modalities has raised a critical question: can these ``jack-of-all-trades'' systems perform on par with highly specialized models in knowledge-intensive domains? This work investigates this question within the high-stakes field of medical image segmentation. We conduct a comparative study analyzing the zero-shot performance of a state-of-the-art omnimodel (Gemini 2.5 Pro, the ``Nano Banana'' model) against domain-specific deep learning models on three distinct tasks: polyp (endoscopy), retinal vessel (fundus), and breast tumor segmentation (ultrasound). Our study focuses on performance at the extremes by curating subsets of the ``easiest'' and ``hardest'' cases based on the specialist models' accuracy. Our findings reveal a nuanced and task-dependent landscape. For polyp and breast tumor segmentation, specialist models excel on easy samples, but the omnimodel demonstrates greater robustness on hard samples where specialists fail catastrophically. Conversely, for the fine-grained task of retinal vessel segmentation, the specialist model maintains superior performance across both easy and hard cases. Intriguingly, qualitative analysis suggests omnimodels may possess higher sensitivity, identifying subtle anatomical features missed by human annotators. Our results indicate that while current omnimodels are not yet a universal replacement for specialists, their unique strengths suggest a potential complementary role with specialist models, particularly in enhancing robustness on challenging edge cases.

We describe a publicly available multimodal dataset of annotated Positron Emission Tomography/Computed Tomography (PET/CT) studies for head and neck cancer research. The dataset includes 1123 FDG-PET/CT studies from patients with histologically confirmed head and neck cancer, acquired from 10 international medical centers. All examinations consisted of co-registered PET/CT scans with varying acquisition protocols, reflecting real-world clinical diversity across institutions. Primary gross tumor volumes (GTVp) and involved lymph nodes (GTVn) were manually segmented by experienced radiation oncologists and radiologists following standardized guidelines and quality control measures. We provide anonymized NifTi files of all studies, along with expert-annotated segmentation masks, radiotherapy dose distribution for a subset of patients, and comprehensive clinical metadata. This metadata includes TNM staging, HPV status, demographics (age and gender), long-term follow-up outcomes, survival times, censoring indicators, and treatment information. We demonstrate how this dataset can be used for three key clinical tasks: automated tumor segmentation, recurrence-free survival prediction, and HPV status classification, providing benchmark results using state-of-the-art deep learning models, including UNet, SegResNet, and multimodal prognostic frameworks.

Recent learning-based approaches have made astonishing advances in calibrated medical imaging like computerized tomography (CT), yet they struggle to generalize in uncalibrated modalities -- notably magnetic resonance (MR) imaging, where performance is highly sensitive to the differences in MR contrast, resolution, and orientation. This prevents broad applicability to diverse real-world clinical protocols. Here we introduce BrainFM, a modality-agnostic, multi-task vision foundation model for human brain imaging. With the proposed "mild-to-severe" intra-subject generation and "real-synth" mix-up training strategy, BrainFM is resilient to the appearance of acquired images (e.g., modality, contrast, deformation, resolution, artifacts), and can be directly applied to five fundamental brain imaging tasks, including image synthesis for CT and T1w/T2w/FLAIR MRI, anatomy segmentation, scalp-to-cortical distance, bias field estimation, and registration. We evaluate the efficacy of BrainFM on eleven public datasets, and demonstrate its robustness and effectiveness across all tasks and input modalities. Code is available at https://github.com/jhuldr/BrainFM.

Delta biomarkers that reflect changes in tumour burden over time can support personalised follow-up in head and neck cancer. However, their clinical use can be limited by the need for manual image segmentation. This study externally evaluates a deep learning model for automatic determination of volume change from serial 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) scans to stratify patients by loco-regional outcome. Patient/material and methods: An externally developed deep learning algorithm for tumour segmentation was applied to pre- and post-radiotherapy (RT, with or without concomitant chemoradiotherapy) PET/CT scans of 50 consecutive head and neck cancer patients from The Christie NHS Foundation Trust, UK. The model, originally trained on pre-treatment scans from a different institution, was deployed to derive tumour volumes at both time points. The AI-derived change in tumour volume (ΔPET-Gross tumour volume (GTV)) was calculated for each patient. Kaplan-Meier analysis assessed loco-regional control based on ΔPET-GTV, dichotomised at the cohort median. In a separate secondary analysis confined to the pre‑treatment scans, a radiation oncologist qualitatively evaluated the AI‑generated PET‑GTV contours. Patients with higher ΔPET-GTV (i.e. greater tumour shrinkage) had significantly improved loco-regional control (log-rank p = 0.02). At 2 years, control was 94.1% (95% CI: 83.6-100%) vs. 53.6% (95% CI: 32.2-89.1%). Only one of nine failures occurred in the high ΔPET-GTV group. Clinician review found AI volumes acceptable for planning in 78% of cases. In two cases, the algorithm identified oropharyngeal primaries on pre-treatment PET-CT before clinical identification. Deep learning-derived ΔPET-GTV may support clinically meaningful assessment of post-treatment disease status and risk stratification, offering a scalable alternative to manual segmentation in PET/CT follow-up.